eeColor Processor - CalMAN - 3D LUTs - Page 16 - AVS Forum
First ... 14  15  16 17  18  ... Last
Display Calibration > eeColor Processor - CalMAN - 3D LUTs
Iron Mike's Avatar Iron Mike 02:13 AM 04-30-2014
Quote:
Originally Posted by sotti View Post

created a set with just those two lines and I don't get that issue.



Make sure the colorset line look something like:
Code:
<ColorSet Name="test" ReadOnly="false" RoundRGB="false" DynamicGamma="false">
if RoundRGB="true", then 222 would be the expected result for 000727 as it would round to 87% flat before going back to bit.

I can't explain why 74.9% would go to 189, as that rounds to 75% which evaluate to 191.

RoundRGB was set to false and ReadOnly was set to false as well... just tested with ReadOnly true... same thing... I'm gonna email you the set in a sec...

ACappo's Avatar ACappo 08:12 AM 04-30-2014
I emailed SpectraCal but was wondering if others have had the same problem?

I have been trying to calibrate my eecolor box but have found the following issue using the latest beta.5.3.5.1551

I did the following to verify my PC HDMI output to my bluray player:

Check clipping levels (WTW & BTB) via flashing bars (OK)
Measure MML and Peak white from both PC and Bluray player (OK)
Measure Color Checker SG from both PC and Bluray player (OK)
Measure 21pt Grayscale from both PC and Bluray player (OK)
Measure 10% increments Saturation Sweep from both PC and Bluray player (OK)
Measure 10% increments Luminance Sweep from both PC and Bluray player (OK)

based on the above im happy with the output from my laptop

I the did a cube calibration using Calman internal patterns
using the detailed option

when done I got the following result (again using Calmans internal patterns)




any idea what could be the problem?

red is slightly undersaturated (uncalibrated) but not to the extend shown above
sotti's Avatar sotti 08:52 AM 04-30-2014
The 5.3.5.1551 beta has a bug that if you are underluminance for a 109%(100% if limited to reference white) we push luminance up by undersaturating the color. This does get the light output correct, but was not our intended effect.

We should be putting up a fix shortly.
ACappo's Avatar ACappo 09:06 AM 04-30-2014
But is this not more a hue issue rather then a saturation problem?
ConnecTEDDD's Avatar ConnecTEDDD 11:58 AM 04-30-2014
CalMAN 5.3.5 Build 1556 Open Beta is available for download.

Release Notes

We've resolved an issue where CalMAN would under saturate colors during a cube calibration in an effort to reach target luminance.
ACappo's Avatar ACappo 11:05 PM 04-30-2014
I have tried with the latest update and got slightly better results smile.gif

Pre Calibration






post calibration using 5.3.5.1556




ACappo's Avatar ACappo 11:16 PM 04-30-2014
I have also tried with reference white only ticked and got very good results (note this is from the beta without the luminance fix)



only problem was introducion of what I would call a slight banding, best illustrated by the spectrum test patterns from Ted's disk. when no lut/unity is applied there is a smooth transaction between all colors, however with the lut all edges is more "sharp" (its very hard to capture on a picture however let me know if it would help to understand what I mean.
ConnecTEDDD's Avatar ConnecTEDDD 01:11 AM 05-01-2014
Quote:
Originally Posted by ACappo View Post

only problem was introducion of what I would call a slight banding, best illustrated by the spectrum test patterns from Ted's disk. when no lut/unity is applied there is a smooth transaction between all colors, however with the lut all edges is more "sharp" (its very hard to capture on a picture however let me know if it would help to understand what I mean.

Hi, what display are you trying to calibrate?
ACappo's Avatar ACappo 02:19 AM 05-01-2014
the test display is an older samsung LCD LN46A650
ConnecTEDDD's Avatar ConnecTEDDD 02:23 AM 05-01-2014
Quote:
Originally Posted by ACappo View Post

the test display is an older samsung LCD LN46A650

Try CalMAN 5.2.3.1416
PE06MCG's Avatar PE06MCG 04:07 AM 05-01-2014
Quote:
Originally Posted by ConnecTEDDD View Post

Try CalMAN 5.2.3.1416

Not specifically CM but including all Calibration software I have a question.

The pursuit of more and more chart accuracy (which may or may not be visible) is admirable but a concern is that perhaps the software used is unintentionally introducing other PQ defects that are visible on some types / makes / models of Displays.
Is this possible ?

Ted's post #45 on the webpage below indicates that the overuse of some TV controls can cause visible problems of the type I mean.
Perhaps calibration software may do the same and should be used with similar discretion on certain Hardware ?

http://www.avsforum.com/t/1529378/satisfaction-with-3d-lut-calibrations/30
zoyd's Avatar zoyd 07:34 AM 05-01-2014
Quote:
Originally Posted by ACappo View Post


only problem was introducion of what I would call a slight banding, best illustrated by the spectrum test patterns from Ted's disk. when no lut/unity is applied there is a smooth transaction between all colors, however with the lut all edges is more "sharp" (its very hard to capture on a picture however let me know if it would help to understand what I mean.

Can you check using the 12 color 1-step patterns which color the banding is most noticeable?

I'm assuming you are playing back the disk in YCbCr, if you can switch to RGB limited do they go away?
ACappo's Avatar ACappo 09:16 AM 05-01-2014
Zoyd,

Not sure what this mean "Can you check using the 12 color 1-step patterns which color the banding is most noticeable?" smile.gif

regarding the disk play back, yes my player output YCbCr I'm not sure it will output RGB but I will test tonight.
zoyd's Avatar zoyd 10:10 AM 05-01-2014
Quote:
Originally Posted by ACappo View Post

Zoyd,

Not sure what this mean "Can you check using the 12 color 1-step patterns which color the banding is most noticeable?" smile.gif

Check for banding in Ted's 1-step 12 color charts, I think it's the page before the spectrum test.
fight4yu's Avatar fight4yu 12:01 PM 05-01-2014
Quote:
Originally Posted by fight4yu View Post

Derek,

Question for CM on DLC. in the report, it mentioned it only have to set brightness/contrast, and then at 100% WHITE set to D65. Is this ALL I have to do before running 3D CUBE?
After the 3D CUBE, do I still need to run grayscale, or i assume CM already take care of everything?

If it does, then 4.5 hr for DLC is not too bad, as currently today, grayscale take somewhat like 30 mins and 3D CUBE about an hr for just a 5x5x5. so, i am already spending 1.5 hrs for 21 + 125 pts. If eecolor takes 4.5 hrs TOTAL and I can have better accuracy, this is not too bad. 5 hr (including setup and etc) is still pretty long, but I assume the more linear the display, the faster it will be??

A big win for current CALMAN owner. Thanks.

Don't think this had been answered... but do I still need to run grayscale or is autocube with eecolor takes care of everything?

Thanks.
buzzard767's Avatar buzzard767 12:24 PM 05-01-2014
Quote:
Originally Posted by fight4yu View Post

Quote:
Originally Posted by fight4yu View Post

Derek,

Question for CM on DLC. in the report, it mentioned it only have to set brightness/contrast, and then at 100% WHITE set to D65. Is this ALL I have to do before running 3D CUBE?
After the 3D CUBE, do I still need to run grayscale, or i assume CM already take care of everything?

If it does, then 4.5 hr for DLC is not too bad, as currently today, grayscale take somewhat like 30 mins and 3D CUBE about an hr for just a 5x5x5. so, i am already spending 1.5 hrs for 21 + 125 pts. If eecolor takes 4.5 hrs TOTAL and I can have better accuracy, this is not too bad. 5 hr (including setup and etc) is still pretty long, but I assume the more linear the display, the faster it will be??

A big win for current CALMAN owner. Thanks.

Don't think this had been answered... but do I still need to run grayscale or is autocube with eecolor takes care of everything?

Thanks.

Check the note at the top of the Calibrate RGB Balance module.
ACappo's Avatar ACappo 01:00 PM 05-01-2014
Quote:
Originally Posted by derekjsmith View Post

I will get our eeColor/ColorBox tool uploaded this week. It has support for updating the OSD images from BMP files.

Not sure if i missed this but was it ever uploaded?
ConnecTEDDD's Avatar ConnecTEDDD 02:24 PM 05-01-2014
Quote:
Originally Posted by zoyd View Post

Quote:
Originally Posted by ACappo View Post

Zoyd,

Not sure what this mean "Can you check using the 12 color 1-step patterns which color the banding is most noticeable?" smile.gif

Check for banding in Ted's 1-step 12 color charts, I think it's the page before the spectrum test.



You can use many patterns from that page to identify many type of errors that can't be reported by classic dE reporting.
ConnecTEDDD's Avatar ConnecTEDDD 02:29 PM 05-01-2014
Quote:
Originally Posted by fight4yu View Post

Quote:
Originally Posted by fight4yu View Post

Derek,

Question for CM on DLC. in the report, it mentioned it only have to set brightness/contrast, and then at 100% WHITE set to D65. Is this ALL I have to do before running 3D CUBE?
After the 3D CUBE, do I still need to run grayscale, or i assume CM already take care of everything?

If it does, then 4.5 hr for DLC is not too bad, as currently today, grayscale take somewhat like 30 mins and 3D CUBE about an hr for just a 5x5x5. so, i am already spending 1.5 hrs for 21 + 125 pts. If eecolor takes 4.5 hrs TOTAL and I can have better accuracy, this is not too bad. 5 hr (including setup and etc) is still pretty long, but I assume the more linear the display, the faster it will be??

A big win for current CALMAN owner. Thanks.

Don't think this had been answered... but do I still need to run grayscale or is autocube with eecolor takes care of everything?

Thanks.

Hi, after the AutoCAL you don't need to change anything, 65-Point Grayscale is included inside the 65-Point 3D LUT Table.

At previous versions of CalMAN, SpectaCAL was recommended 31-Point Grayscale + 31-Point RGBCMY Luminance combined with 9-Point Cube for eeColor.

Now with the latest version it has to do with the display....CalMAN is deciding how many color points to use for your specific display/projector settings.
sotti's Avatar sotti 03:42 PM 05-01-2014
With the Grayscale it can help the AutoCube process run more quickly if you do the 1D LUT first as we do have more control over grayscale performance with the 1D LUTs as each point is a direct look up, where a 3D LUT has points that are dependent on interpolation, and trilinear interpolation can suffer from artifacts if we don't have enough points.

At 64/65 points, we typically have sufficient density that the end result won't be significantly different, but like I mentioned, it may be a bit fast with the grayscale calibrated in advance.
ConnecTEDDD's Avatar ConnecTEDDD 03:56 PM 05-01-2014
Quote:
Originally Posted by sotti View Post

With the Grayscale it can help the AutoCube process run more quickly if you do the 1D LUT first as we do have more control over grayscale performance with the 1D LUTs as each point is a direct look up, where a 3D LUT has points that are dependent on interpolation, and trilinear interpolation can suffer from artifacts if we don't have enough points.

At 64/65 points, we typically have sufficient density that the end result won't be significantly different, but like I mentioned, it may be a bit fast with the grayscale calibrated in advance.

Yes, to do Grayscale using your display internal controls is recommended and it will provide you better find results in case that your RGB Balance controls are working and not introducing problems to your image when you are tweaking them.

I'm suggesting also to do the Grayscale first using your display controls and later load some Color Reproduction Patterns before starting the AutoCAL.

Check using ramp color ramps/multi color-steps.....if you see any problem with your current display control settings you have used, then try to re-do your Grayscale by using less tweaking of controls this time, less changes as possible to your RGB Gain/Cut sliders....

Later re-check the Color Reproduction Patterns again... If all good, then move to 3D LUT AutoCal with eeColor.
ACappo's Avatar ACappo 02:52 PM 05-02-2014
Quote:
Originally Posted by zoyd View Post

Can you check using the 12 color 1-step patterns which color the banding is most noticeable?

I'm assuming you are playing back the disk in YCbCr, if you can switch to RGB limited do they go away?

wow. I did finally have a chance to test with the disk (based on the latest LUT (as in post 456) I have unfortunately overwritten post 457's LUT
and I must say the result is not pretty.

looking at color ramp 1 & 2 its easy to see the banding at


The luminance steps looked OK, just wondered if im supposed to see the square around the squares

This is of the 21 step saturation bars im surprised to see issues here since when measuring the saturation sweeps it looked good


This is of the saturation ramps


any comments would be great (particular from Spectracal:) )
ConnecTEDDD's Avatar ConnecTEDDD 03:10 PM 05-02-2014
Hello, there a lot of problems to this visual verification of the cLUT that CalMAN generated.

The square outline is probably coming from your sharpness setting, please check this red pattern again with UNITY LUT Active, if the outline is still there, load the Advanced Sharpness Pattern and tweak your sharpness setting.

There is a lot of noise to your Saturation Ramps Picture, can you re-capture these photos with a UNITY LUT active?

Your pictures are showcase exactly that there a lot of things that the dE reports/charts can't show, that's why I designed so many patterns for verification of the generated cLUT.

Is your source your PC and you are using your HDMI Output? Do you playback my disk from your PC Player, right?
zoyd's Avatar zoyd 03:17 PM 05-02-2014
Quote:
Originally Posted by ACappo View Post

wow. I did finally have a chance to test with the disk (based on the latest LUT (as in post 456) I have unfortunately overwritten post 457's LUT
and I must say the result is not pretty.

That is what I am seeing as well with YCC playback, RGB limited and full scale is smooth. Spectracal is aware of this issue.
ACappo's Avatar ACappo 03:36 PM 05-02-2014
so since my bluray player will not output RGB there is currently nothing I can do?
Iron Mike's Avatar Iron Mike 03:41 PM 05-02-2014
Quote:
Originally Posted by Iron Mike View Post

RoundRGB was set to false and ReadOnly was set to false as well... just tested with ReadOnly true... same thing... I'm gonna email you the set in a sec...

Joel,

did u get that email and can u tell why the RGB values of custom patch sets are incorrectly calculated in CM 5 ?

- M
ACappo's Avatar ACappo 04:43 PM 05-02-2014
just a few notes to the above.

before autocal I verified that clipping levels was identical between PC and bluray player, that MML and peak white was the same fl that sharpness was the same

when I turn off the LUT or use unity lut all ramps are smooth and no issues
I dont have bluray playback software on my laptop so cant verify with teds disk from my laptop. (anyone know of a free player?
Iron Mike's Avatar Iron Mike 04:47 PM 05-02-2014
Quote:
Originally Posted by ACappo View Post

I dont have bluray playback software on my laptop so cant verify with teds disk from my laptop. (anyone know of a free player?

VLC
ACappo's Avatar ACappo 05:46 PM 05-02-2014
Thank you Mike.

after playing the disk from my PC, I can confirm the issues are also there when played from my PC.
ConnecTEDDD's Avatar ConnecTEDDD 03:29 AM 05-03-2014
Quote:
Originally Posted by ACappo View Post

Thank you Mike.

after playing the disk from my PC, I can confirm the issues are also there when played from my PC.

Hi, If you have time try again a new AutoCAL using an official released version of CalMAN and not a Beta Build.
First ... 14  15  16 17  18  ... Last

Up
Mobile  Desktop